Developing a Quantitative Ultrasound Image Feature Analysis Scheme to Assess Tumor Treatment Efficacy Using a Mouse Model

A despeckling method for ultrasound images utilizing content-aware prior and attention-driven techniques

The despeckling of ultrasound images contributes to the enhancement of image quality and facilitates precise treatment of conditions such as tumor cancers. However, the use of existing methods for eliminating speckle noise can cause the loss of image texture features, impacting clinical judgment. Thus, maintaining clear lesion boundaries while eliminating speckle noise is a challenging task. This paper presents an innovative approach for denoising ultrasound images using a novel noise reduction network model called content-aware prior and attention-driven (CAPAD). The model employs a neural network to automatically capture the hidden prior features in ultrasound images to guide denoising and embeds the denoiser into the optimization module to simultaneously optimize parameters and noise. Moreover, this model incorporates a content-aware attention module and a loss function that preserves the structural characteristics of the image. These additions enhance the network's capacity to capture and retain valuable information. Extensive qualitative evaluation and quantitative analysis performed on a comprehensive dataset provide compelling evidence of the model's superior denoising capabilities. It excels in noise suppression while successfully preserving the underlying structures within the ultrasound images. Compared to other denoising algorithms, it demonstrates an improvement of approximately 5.88% in PSNR and approximately 3.61% in SSIM. Furthermore, using CAPAD as a preprocessing step for breast tumor segmentation in ultrasound images can greatly improve the accuracy of image segmentation. The experimental results indicate that the utilization of CAPAD leads to a notable enhancement of 10.43% in the AUPRC for breast cancer tumor segmentation.

Image-guided intratumoral immunotherapy: Developing a clinically practical technology

Immunotherapy has revolutionized the contemporary oncology landscape, with durable responses possible across a range of cancer types. However, the majority of cancer patients do not respond to immunotherapy due to numerous immunosuppressive barriers. Efforts to overcome these barriers and increase systemic immunotherapy efficacy have sparked interest in the local intratumoral delivery of immune stimulants to activate the local immune response and subsequently drive systemic tumor immunity. While clinical evaluation of many therapeutic candidates is ongoing, development is hindered by a lack of imaging confirmation of local delivery, insufficient intratumoral drug distribution, and a need for repeated injections. The use of polymeric drug delivery systems, which have been widely used as platforms for both image guidance and controlled drug release, holds promise for delivery of intratumoral immunoadjuvants and the development of an in situ cancer vaccine for patients with metastatic cancer. In this review, we explore the current state of the field for intratumoral delivery and methods for optimizing controlled drug release, as well as practical considerations for drug delivery design to be optimized for clinical image guided delivery particularly by CT and ultrasound.

A multilayer network-enabled ultrasonic image series analysis approach for online cancer drug delivery monitoring

The objective of this study is to develop an effective data-driven methodology for the online monitoring of cancer drug delivery guided by the ultrasonic images. To achieve this goal, effective image quantification and accurate feature extraction play a critical role on image-guided drug delivery (IGDD) monitoring. However, the existing image-guided approaches in such area are mainly focused on the analysis for individual images rather than the image series. In fact, the temporal patterns between consecutive images may contain critical information and it is necessary to be considered in the monitoring analysis. In addition, the conventional approaches, such as the pure intensity-based method, also do not sufficiently consider the effects of noise in the ultrasonic images, which also limits the monitoring sensitivity and accuracy. To address the challenges, this paper proposed a novel multilayer network-enabled IGDD (MNE-IGDD) monitoring approach. The contributions of the proposed method can be summarized into three aspects: (1) formulate the sequential ultrasound images to a multilayer network by the proposed spatial-regularized distance; (2) detect drug delivery area based on community detection algorithm of multilayer network; and (3) quantify the drug delivery progress by incorporating the image intensity-based features with the detected community. Both the detected communities and feature increment percentages are applied as the evaluation metric for validation. A simulation study was conducted and this method was also applied to a real-world mouse colon tumor treatment case study under three temperature conditions. Both simulation and the real-world case studies demonstrated that the proposed method is promising to achieve satisfactory monitoring performance in clinical trials.

Bibliometric Analysis of Global Research Trends on Ultrasound Microbubble: A Quickly Developing Field

Background: Microbubbles are widely used as highly effective contrast agents to improve the diagnostic capability of ultrasound imaging. Mounting evidence suggests that ultrasound coupled with microbubbles has promising therapeutic applications in cancer, cardiovascular, and neurological disorders by acting as gene or drug carriers. The aim of this study was to identify the scientific output and activity related to ultrasound microbubble through bibliometric approaches. Methods: The literature related to ultrasound microbubble published between 1998 and 2019 was identified and selected from the Science Citation Index Expanded of Web of Science Core Collection on February 21, 2021. The Scopus database was also searched to validate the results and provided as supplementary material. Quantitative variables including number of publications and citations, H-index, and journal citation reports were analyzed by using Microsoft Excel 2019 and GraphPad Prism 8.0 software. VOS viewer and CiteSpace V were used to perform coauthorship, citation, co-citation, and co-occurrence analysis for countries/regions, institutions, authors, and keywords. Results: A total of 6088 publications from the WoSCC were included. The United States has made the largest contribution in this field, with the majority of publications (2090, 34.3%), citations (90,741, 46.6%), the highest H-index (138), and close collaborations with China and Canada. The most contributive institution was the University of Toronto. Professors De Jong N and Dayton P A have made great achievements in this field. However, the research cooperation between institutions and authors was relatively weak. All the studies could be divided into four clusters: "ultrasound diagnosis study," "microbubbles' characteristics study," "gene therapy study," and "drug delivery study." The average appearing years (AAY) of keywords in the cluster "drug delivery study" was more recent than other clusters. For promising hot spots, "doxorubicin" showed a relatively latest AAY of 2015.49, followed by "nanoparticles" and "breast cancer." Conclusion: There has been an increasing amount of scientific output on ultrasound microbubble according to the global trends, and the United States is staying ahead in this field. Collaboration between research teams still needs to be strengthened. The focus gradually shifts from "ultrasound diagnosis study" to "drug delivery study." It is recommended to pay attention to the latest hot spots, such as "doxorubicin," "nanoparticles," and "breast cancer."

Applying a random projection algorithm to optimize machine learning model for predicting peritoneal metastasis in gastric cancer patients using CT images

BACKGROUND AND OBJECTIVE

Non-invasively predicting the risk of cancer metastasis before surgery can play an essential role in determining which patients can benefit from neoadjuvant chemotherapy. This study aims to investigate and test the advantages of applying a random projection algorithm to develop and optimize a radiomics-based machine learning model to predict peritoneal metastasis in gastric cancer patients using a small and imbalanced computed tomography (CT) image dataset.

METHODS

A retrospective dataset involving CT images acquired from 159 patients is assembled, including 121 and 38 cases with and without peritoneal metastasis, respectively. A computer-aided detection scheme is first applied to segment primary gastric tumor volumes and initially compute 315 image features. Then, five gradients boosting machine (GBM) models embedded with five feature selection methods (including random projection algorithm, principal component analysis, least absolute shrinkage, and selection operator, maximum relevance and minimum redundancy, and recursive feature elimination) along with a synthetic minority oversampling technique, are built to predict the risk of peritoneal metastasis. All GBM models are trained and tested using a leave-one-case-out cross-validation method.

RESULTS

Results show that the GBM model embedded with a random projection algorithm yields a significantly higher prediction accuracy (71.2%) than the other four GBM models (p<0.05). The precision, sensitivity, and specificity of this optimal GBM model are 65.78%, 43.10%, and 87.12%, respectively.

CONCLUSIONS

This study demonstrates that CT images of the primary gastric tumors contain discriminatory information to predict the risk of peritoneal metastasis, and a random projection algorithm is a promising method to generate optimal feature vector, improving the performance of machine learning based prediction models.

Establishment of an Endoscopy-Guided Minimally Invasive Orthotopic Mouse Model of Colorectal Cancer

Simple Summary

Open orthotopic mouse models of colorectal cancer have disadvantages such as the requirement for advanced surgical skills or the trauma caused by laparotomy. To overcome these limitations, this study aimed to evaluate the establishment of an endoscopy-guided minimally invasive model without laparotomy. Different concentrations of the murine CRC cell lines CT26 and MC38 were endoscopically injected into the colorectal wall of BALB/C and C57BL/6J mice, respectively. Consistent tumor growth with the presence of tumor-infiltrating lymphocytes, lympho-vascular invasion, and early spontaneous lymph node, peritoneal, and hepatic metastases were observed. Analysis of the learning curve demonstrated that this model is easy to learn and quick to establish. It enables intra-individual follow-up endoscopies, and features tumors to study mechanisms of metastasis and the interaction with the immune system. The application of specific cell lines and concentrations enables a controlled local tumor growth and metastatic formation within short observation periods.

Abstract

Open orthotopic mouse models of colorectal cancer have disadvantages such as the requirement for advanced surgical skills or the trauma caused by laparotomy. To overcome these drawbacks, this study aimed to evaluate the establishment of a minimally invasive model using murine colonoscopy. CT26 and MC38 CRC cells of different concentrations were injected into BALB/C and C57BL/6J mice, respectively. Follow-up endoscopies were performed to assign an endoscopic score to tumor growth. Gross autopsy, histologic and immuno-histochemical evaluation, and immune scoring were performed. To describe the learning curve of the procedures, a performance score was given. Local tumor growth with colorectal wall infiltration, luminal ulceration, the presence of tumor-infiltrating lymphocytes, lympho-vascular invasion, and early spontaneous lymph node, peritoneal, and hepatic metastases were observed. The tumors showed cytoplasmic immuno-staining for CK20. Compared to the MC38/C57BL/6J model, tumorigenicity and immunogenicity of the CT26/BALB/C model were higher. Tumor volume correlated with the endoscopic score. This endoscopy-guided orthotopic mouse model is easy to learn and quick to establish. It features early metastasis and enables the study of interactions with the immune system. When specific cell concentrations and cell lines are applied, controlled local tumor growth and metastasis can be achieved within short observation periods.

Establishment of a guided, in vivo, multi-channel, abdominal, tissue imaging approach

Novel tools in humane animal research should benefit the animal as well as the experimentally obtained data. Imaging technologies have proven to be versatile and also in accordance with the demands of the 3 R principle. However, most imaging technologies are either limited by the target organs, number of repetitive imaging sessions, or the maximal resolution. We present a technique-, which enables multicolor abdominal imaging on a tissue level. It is based on a small imaging fiber endoscope, which is guided by a second commercial endoscope. The imaging fiber endoscope allows the distinction of four different fluorescence channels. It has a size of less than 1 mm and can approximately resolve single cells. The imaging fiber was successfully tested on cells in vitro, excised organ tissue, and in mice in vivo. Combined with neural networks for image restauration, high quality images from various abdominal organs of interest were realized. The second endoscope ensured a precise placement of the imaging fiber in vivo. Our approach of guided tissue imaging in vivo, combined with neuronal networks for image restauration, permits the acquisition of fluorescence-microscope like images with minimal invasive surgery in vivo. Therefore, it is possible to extend our approach to repetitive imaging sessions. The cost below 30 thousand euros allows an establishment of this approach in various scenarios.

Developing global image feature analysis models to predict cancer risk and prognosis

In order to develop precision or personalized medicine, identifying new quantitative imaging markers and building machine learning models to predict cancer risk and prognosis has been attracting broad research interest recently. Most of these research approaches use the similar concepts of the conventional computer-aided detection schemes of medical images, which include steps in detecting and segmenting suspicious regions or tumors, followed by training machine learning models based on the fusion of multiple image features computed from the segmented regions or tumors. However, due to the heterogeneity and boundary fuzziness of the suspicious regions or tumors, segmenting subtle regions is often difficult and unreliable. Additionally, ignoring global and/or background parenchymal tissue characteristics may also be a limitation of the conventional approaches. In our recent studies, we investigated the feasibility of developing new computer-aided schemes implemented with the machine learning models that are trained by global image features to predict cancer risk and prognosis. We trained and tested several models using images obtained from full-field digital mammography, magnetic resonance imaging, and computed tomography of breast, lung, and ovarian cancers. Study results showed that many of these new models yielded higher performance than other approaches used in current clinical practice. Furthermore, the computed global image features also contain complementary information from the features computed from the segmented regions or tumors in predicting cancer prognosis. Therefore, the global image features can be used alone to develop new case-based prediction models or can be added to current tumor-based models to increase their discriminatory power.